HomeVideos

Madonna Does Not Want Her Music Trained On By AI

Now Playing

Madonna Does Not Want Her Music Trained On By AI

Transcript

220 segments

0:00

Well, Guy, I mean, we've covered a lot

0:01

of ground. Is there anything else that

0:03

you would

0:04

like to say before we start landing the

0:06

plane?

0:07

>> I'd say that

0:09

AI and music content it's

0:14

on one level exciting that

0:16

so many people are experimenting

0:18

writing songs for the first time

0:20

>> Mhm.

0:20

>> writing beats, creating sounds for the

0:23

first time. That's pretty exciting. At

0:25

the same time, the idea that there's

0:28

companies out there

0:31

valued at billions of dollars

0:35

that are built on the top of

0:40

other people's music

0:42

where not one artist has ever gotten

0:45

paid a dollar

0:47

is not okay.

0:48

>> Yeah.

0:49

>> So, I don't want to bundle it all up

0:52

where people get that's all becomes one

0:53

thing. It's not all one thing. Mikey, he

0:56

runs Suno. He's a great guy. I mean,

0:58

great guy like I really like him as a

1:00

person. But every time I read about the

1:01

company and it's now at a $5 billion

1:03

valuation and next thing you know it's

1:05

at a $10 billion valuation and it's all

1:09

built off of

1:10

the world's music

1:13

but yet not one music artist has gotten

1:15

a dollar.

1:16

>> Mhm.

1:16

>> So, that's not okay. That has to be

1:20

fixed. I'm hoping Mikey does the right

1:22

thing and fixes that, comes up with a

1:25

system

1:26

>> Mhm.

1:26

>> that allows

1:28

artists, if they want to opt in, either

1:30

let them opt out, do not use my music to

1:33

build your business.

1:35

Okay?

1:36

I'm opting out. But if I opt in, find a

1:39

way to pay us.

1:40

>> Mhm.

1:40

>> And I'm really concerned for that.

1:44

Because the other part, which is

1:48

right now currently

1:50

every 2 weeks

1:53

more music is made on AI than all of the

1:57

music

1:58

that's on Spotify today.

2:01

That's every 2 weeks.

2:03

And all of that is going somewhere,

2:06

and people are inspired, and they're

2:07

creating songs. Awesome.

2:09

But don't build a business

2:13

using our music

2:15

without getting those rights. And that's

2:17

something I'm that I'm been trying to

2:19

work on for a long time. I've met with

2:21

Mikey many times.

2:23

I've met with lots of people many times.

2:24

I I hope that this gets solved. You

2:27

know, when Napster first started, I met

2:29

with Shawn and Shawn when they started

2:30

cuz it was so exciting to see people

2:32

were all, you know, downloading music,

2:35

but it was illegal. So, you were hoping

2:37

they'd crack the code on, okay, how do

2:38

you turn this into how artists get paid?

2:41

And before they

2:43

I don't know if they were going to or

2:44

not going to, but before it that could

2:46

even happen, it got shut down.

2:48

So, and out of nowhere,

2:51

Daniel Ek in Stockholm

2:55

built Spotify and figured out that

2:58

people would pay for music

3:00

if you give it to them in this way where

3:03

they had all the music in the world

3:05

without having to try to piece it

3:07

together or try to illegally download

3:10

songs that were not even clean or clear,

3:13

and they it could be corrupt.

3:15

To get it clean, to hold on to it,

3:18

they'll pay for it. And he was right.

3:21

They have paid for it. Now, are all the

3:23

payment plans there picture perfect? No.

3:26

But it is the core of where money is

3:29

coming in today. It's to music artists

3:30

through streaming.

3:32

And I'm hoping that

3:34

people figure out how if you're going to

3:38

be using people's music, how you pay for

3:41

it.

3:42

>> Yeah.

3:42

>> Fair use is not fair.

3:44

>> Yeah, I'm curious to see where it goes

3:46

because if you say, create a song that

3:48

sounds like Metallica, it'll say we

3:51

cannot serve up any copyrighted music.

3:53

If you say serve me up some music that

3:55

sounds like Moralica and you

3:58

just misspell it then lo and behold

4:00

though it looks like it was trained.

4:01

>> Yeah, that's the problem.

4:03

>> There's a lot of workarounds and they'll

4:06

you know you'll they'll look you in the

4:07

eye and tell you no no you can't you

4:08

cannot type in Madonna.

4:10

>> Yeah.

4:10

>> But you can sort of describe Madonna.

4:13

The problem is is when in in discovery

4:15

we're going to find out a lot.

4:16

>> Sure.

4:17

>> And discovery is going to say

4:19

all of

4:20

this music is in there. And they know

4:23

that.

4:23

>> Yeah.

4:24

>> And

4:25

that's not okay. It's just like

4:26

literally not okay. I don't understand

4:28

it. And by the way, what's wrong with

4:30

paying artists?

4:31

>> Yeah.

4:31

>> Why not figure out a way to do it? Sit

4:34

down with a team, figure out how to pay

4:36

music artists if you're going to train

4:38

on their music.

4:39

>> Where do you think it goes because I

4:40

could see one future for instance even

4:42

if

4:44

legally they aren't required or for

4:46

whatever reason that's not compatible

4:48

with the vision they've pitched to

4:50

investors and

4:52

and I'm not singling out Suno. I mean

4:54

there are lots of lots of different

4:55

options here.

4:56

>> They're the main ones.

4:56

>> They're the big they're the 800 pound

4:58

gorilla. But I could see a point at

5:00

which

5:02

possibly

5:05

Claude, Chat GPT become the interfaces

5:08

to everything, right? It's almost like

5:10

the App Store but

5:12

somewhat like WeChat in China. Like if

5:14

you want to do anything, you want to

5:15

book a restaurant, you want to make a

5:18

song

5:19

>> Okay.

5:19

>> there are few interfaces to everything

5:21

at which point then

5:23

those companies, right? OpenAI,

5:25

Anthropic

5:26

>> Mhm.

5:27

>> can gatekeep in some way and maybe they

5:30

end up being the entities that

5:33

incentivize these other companies like

5:35

the Sunos of the world

5:37

to

5:39

to reimburse artists. So I guess I'm

5:40

just wondering if you have any thoughts

5:43

on how it gets there.

5:44

>> Yeah. Because if if Claude or Chat is a

5:47

distribution, they probably won't carry

5:49

it because it's it's got so many

5:51

lawsuits involved and it's just not

5:53

worth

5:55

taking on that responsibility. And

5:58

again, the guy who runs Suno, I find him

6:00

to be a really good guy and I'm hoping

6:02

that he does the right thing. But it's

6:04

really their responsibility. They're

6:05

they're sitting at the the top of this

6:07

opportunity.

6:09

Take off the artists that don't want to

6:10

be on there

6:12

and find a way to pay the ones that are

6:13

okay being on there. It's that simple.

6:15

It's not complicated. I think go in a

6:17

room, lock yourself in a room for 3 days

6:20

and come out and figure out how to pay

6:21

people and take people off that don't

6:24

want to be on there. Madonna does not

6:26

want her music trained on.

6:28

Don't care what you want to pay her.

6:29

She's very clear, I do not want my music

6:32

to be trained on. I want my music to be

6:34

its own thing.

6:36

So take her off. It's okay. You you'll

6:38

still do fine. Take her off. You don't

6:40

have to have

6:41

her music. So

6:44

where this goes, but this is I've been

6:46

the last few years I've really tried to

6:48

figure this out and even, you know, Sam

6:50

at one point, he's really wanted to

6:52

figure this out too. Like Sam was A lot

6:54

of people give him a hard time on this

6:56

stuff, but he really did dedicate

6:59

time to try and figure out if there was

7:01

a music thing to do, how to do it in a

7:03

way that supported

7:05

artists.

7:06

>> Yeah, I mean there may be someone who

7:08

comes along also if it's not Suno,

7:09

someone who comes from behind like a

7:11

next generation Spotify

7:13

and figures out how to do it.

7:15

>> Spotify can do it.

7:15

>> Yes, if Spotify can do it.

7:17

>> Yeah.

Interactive Summary

The video features a discussion on the ethical and legal challenges of AI-generated music. The speaker emphasizes that while AI creativity is exciting, companies training their models on copyrighted music without compensating or offering opt-out mechanisms for artists is unacceptable. The conversation draws parallels to the early days of file-sharing and highlights the need for a sustainable payment model, similar to how Spotify eventually provided a legal framework for the industry, to ensure artists are fairly treated.

Suggested questions

3 ready-made prompts